Digital Object Identifier 10.1109/MMM.2010.937730

Frank Ellinger (Frank.Ellinger@tu-dresden.de), Uwe Mayer, Michael Wickert, Niko Joram, Jens Wagner, and Ralf Eickhoff are with the Technical University of Dresden, Chair for Circuit Design and Network Theory, 01062 Dresden, Germany. Ignacio Santamaria is with

the University of Cantabria, Department of Communications Engineering, Santander, 39005, Spain. Christoph Scheytt and Rolf Kraemer are with IHP Technologies, 15236 Frankfurt (Oder), Germany.

When examining a monthly bank account statement, it is not

only the number below the bottom line that matters. Whether

that number has a minus or plus in front of it is crucial. For

many technical problems, the sign matters as well. In circuits,

we can change the sign by means of phase shifters. Moreover,

by using phase shifters, intermediate states between the signs (including com-

plex values) can be set in circuits. Hence, phase shifters play an important role in

October 2010 97 1527-3342/10/$26.00©2010 IEEE

Frank Ellinger,Uwe Mayer,

Michael Wickert, Niko Joram,Jens Wagner,Ralf Eickhoff,

Ignacio Santamaria,

Christoph Scheytt,and Rolf Kraemer

Integrated Adjustable

Phase Shifters

© FOTOSEARCH

98 October 2010

electrical engineering. Unfortunately, this article does

not give direct insights to change the sign of your bank

statement. However, it aims to give a comprehensive

overview of tunable phase shifters for radio frequen-

cy (RF) applications including cookbooklike design

guidelines and performance comparisons. The focus

of this article is put on phase shifters fully integrated

in a chip.

Phase shifters are required for many RF systems.

Examples are as follows:

Smart antenna systems (e.g., to steer the signal •

beam and to optimally exploit diversity)

Measurement equipment (e.g., for the generation •

and alignment of differential signals)

Amplifier linearization systems (e.g., to cancel •

undesired frequency components)

Phase-locked loops (to align the phase of the •

oscillator signal with that of a reference)

Image rejection receivers (to set the phase such •

that the undesired image can be canceled)

High-speed multirate front ends for optical com- •

munications (by means of multiple phase paths

more sampling points are generated relaxing the

clock speed at same data speed).

RequirementsFor the development and optimization of integrated

phase shifters, a broad range of design constraints

have to be considered. Among the key design param-

eters and requirements are

Phase control range • : frequently 360° to reach the

full phase space

Phase resolution • : theoretically each intermedi-

ate phase can be reached if the control is ana-

logue. In this context, a linear control response

is favorable to minimize the corresponding reso-

lution required for the digital to analogue con-

verters (DACs). The phase control resolution of

switched-type phase shifters is limited by the

available bit resolution

Group delay • : can have a significant impact on

broadband signals

Insertion loss • : gain can be achieved by active

implementations, whereas the dc power saved

in passive circuits can be used for additional

preamplifiers

Variation of insertion loss versus phase • : may require

compensation e.g., by means of a variable gain

amplifier (VGA)

Bandwidth • : the loss, phase, and group delay char-

acteristics versus frequency are important. A

high bandwidth tends to have a lower sensitivity

to process and temperature variations

RF large signal capabilities • : for example, repre-

sented by the 1 dB compression point at the input

(P1dB) and the third-order intercept point at the

input (IIP3)

Power consumption • : relevant especially for bat-

tery-driven devices and for arrays with multiple

paths

Chip size • : as small as possible to decrease costs and

device dimensions, especially for system arrays

Noise • : important for some applications, for exam-

ple in phased array receivers, where the phase

shifters are directly connected with the antennas

System stability • : in advance and from system per-

spective, the designer should carefully take into

account stability at all phase states. In this con-

text, parasitic feedback loops have to be consid-

ered, for example due to substrate coupling and

the nonideal grounds on a printed circuit boards

(PCBs).

This multiplicity of constraints makes the design

of integrated phase shifters to a challenging task in

which reasonable tradeoffs have to be found. The indi-

vidual system requirements have to be considered for

the optimum choice of the phase shifter topology.

TopologiesA variety of RF PCB-based [1] and fully integrated

[2] active and passive phase shifter approaches have

been invented to meet specific requirements. Pas-

sive approaches exhibit insertion loss while provid-

ing good large signal properties. The most common

integrated phase shifters are based on vector modu-

lator, distributed, reflective-type, and switched-type

topologies. Pros and cons of the approaches, includ-

ing subtypes, are summarized in Table 1. In the

following sections, we will discuss the functional

principles of these approaches, including cookbook-

like design guidelines, IC hardware, and perfor-

mance examples.

Vector ModulatorsFigure 1 sketches the architecture of a vector modula-

tor with a phase control range of 90°. The input signal

is divided into two paths with 90° (I/Q) offset, which

are amplitude-weighted by VGAs or attenuators, and

finally combined. Various modifications are possible;

for example, the phase offset may also be implemented

in front of the amplifiers, the amplifiers may act as

dividers or combiners, and the combiner and the phase

offset may be accomplished by means of a 90° hybrid

or low-pass/high-pass filters. LC (inductor capacitor)

filters allow low losses but only small bandwidths,

whereas resistor capacitor (RC)-based topologies

The most common integrated phase shifters are based on vector modulator, distributed, reflective-type, and switched-type topologies.

October 2010 99

enable higher bandwidth at the expense of significant

loss. According to Figure 1(b), the resulting phase and

magnitude are given by

/Vr5 arctanV90°

V0° (1)

and

0Vr 0 5" 0V0°2 0 1 0V90°

2 0 . (2)

Obviously, every phase between 0° and 90° can be

achieved by weighting the path gains. To obtain constant

overall gain, the individual gains of the branches have to

be set properly, for example, by using a look-up table.

Gain/attenuation control components with high

gain control range, typically in the excess of 20 dB, are

TABLE 1. Typical qualitative performance of integrated phase shifter approaches. TL: transmission lines, LP: low-pass filter, HP: high-pass filter, gm: transconductance, Z: impedance, L: inductance.

Type Subtype CommentPhase control Loss/ripple BW Size Noise PDC

Vector modulator

VGA Phase offsets e.g., by filters or couplers

LargeLow (gain)/low

Small/ Medium

Large LowMedium/large

Attenuators Large High/low Medium MediumSimilar to loss

Zero

Distributed

TL Varactor tuned

LargeMedium/ medium

Large Very largeSimilar to loss

Zero

LP-sections

Varactor tuned

LargeMedium/ medium

Large LargeSimilar to loss

Zero

Varactor and active L tuned

Very largeSmall/ medium

Very large Large LargeMedium/ large

Reflective type

Coupler based

Varactor tuned

Small Small/small Medium MediumSimilar to loss

Zero

Varactor tuned, resonated with L

MediumMedium/ large

Small MediumSimilar to loss

Zero

Parallel LC resonances

LargeLarge/large

Small LargeSimilar to loss

Zero

Varactor and active L tuned, resonated

LargeMedium/ medium

Small Small LargeMedium/ large

Circulator based

Z tuning SmallMedium/ medium

Large Small Large Medium

gm tuning Medium Small/small Large Small LargeMedium/ large

gm and Z tuning

Large/ medium

Medium/ medium

Large Small LargeMedium/ large

Switched

TLSimple “digital” control

Large, but resolution limited by number of bits

High/ medium

Medium Very LargeSimilar to loss

Zero

HP/LPHigh/ medium

Small LargeSimilar to loss

Zero

If the amplitude control elements provide a sufficient gain control range, vector modulators can be employed for both phase and gain control.

100 October 2010

required to reach the phase range borders. Due to the

high maximum gain and the high possible attenuation,

cascode amplifiers are well suited. Minimum insertion

phase and port impedance variations versus gain have

to be guaranteed. Otherwise, phase and gain errors

result, which may require calibration. In this context,

optimized VGAs were presented in [3]–[6].

If the amplitude control elements provide a suf-

ficient gain control range, vector modulators can be

employed for both phase and gain control. In this

case, in addition to the relative gain magnitudes, the

absolute values of the gain are set. Vector modulators

require no special devices such as optimized varac-

tors. Hence, vector modulators are well suited for inte-

gration in a variety of common technologies.

Typically, to achieve 360° phase control,four paths, each

exhibiting phase offsets of 90°, are required. At any given

time, only two phase branches are selected by switches.

Hence, the number of VGAs can be reduced to two by

path switching. The four phase paths can, for example, be

generated by using optimized quadrature allpass filters

(QAFs) in a differential configuration [7]. In Figure 2, a cir-

cuit hardware example operating at 5.5 GHz in 0.18 mm

CMOS with 1.5 V3 12 mA dc power, 27 dBm IIP3 at

maximum gain, and 1 mm2 chip area is shown [8]. In

addition to the 360° phase control functionality, this vector

modulator also provides gain control functionality with a

control range from 220 dB to 1.5 dB. A circuit in 0.18 mm

complementary metal–oxide–semiconductor (CMOS)

operating at 15–20 GHz with 360° phase control, 6 6 4 dB

loss, less than 112 mW dc power, and 0.7 mm2 chip area

was published in [9]. A vector modulator at 50–56 GHz

in 90-nm CMOS with 4.9 6 0.9 dB loss, 23 mW dc power

and 0.4 mm2 chip area was reported in [10].

Furthermore, architectures with three parallel paths

exhibiting 120° offset are possible. Referring to [11] and

Figure 3, a circuit example operating at 5–5.5 GHz with

Figure 2. Active vector modulator at 5.5 GHz implemented in 0.18 mm CMOS technology [8]. (a) Circuit architecture, (b) measured complex-plane response achieved by simple, uncalibrated variation of the control voltages, and (c) chip photo, chip size: 1.2 3 0.8 mm2.

–1.5

–1.0

–0.5

0.0

0.5

1.0

1.5

–1.5 –1.0 –0.5 0.0 0.5 1.0 1.5

Im {

S21}

Re {S21}

(b)

(c)

In+

In–

Out

QAF

I

Q

0°180°90°

270°Input−

Input+

Path

Selection

VGAs Combiner

Output

Quadrature

Allpass Filter

+

(a)

Div

ider

Com

bin

er

In

V0°

V90°

Out

90°VGA

VGA 0°

(a)

V0° ↓ V90° ↑

180°

270°

90°: V0° Off, V90° On

0°: V0° On,

V90° Off

(b)

Figure 1. Vector modulator with 90° phase control range [2], VGA: variable gain amplifier. (a) circuit architecture and (b) vector diagram.

The higher the capacitance control ratio and the higher the characteristic length of the equivalent line, the higher the tuning range.

October 2010 101

360° phase control range, 0.6 dB gain, P1dB of 29 dBm,

1.5 V 3 7 mA dc power and 1.3 mm2 chip area imple-

mented in 0.6 mm GaAs metal–semiconductor field-

effect transistor (MESFET) technology was reported. A

simple passive version of such a topology with zero dc

power, 9 dB insertion loss, large P1dB of 16.5 dBm, and

0.5 mm2 chip area was demonstrated in [12]. By com-

bining the control voltages of shunt and series attenua-

tors, providing complementary attenuation the number

of control voltages can be reduced to one [13].

Due to their universal applicability, moderate re-

quirements regarding technologies (e.g., no varactors re-

quired), compact size, and ability for both phase and gain

control, vector modulators are used in a variety of adap-

tive antenna, beamforming, and phased array systems.

For details regarding such smart antenna systems using

vector modulators, the reader is referred to [14]–[19].

Distributed Phase ShiftersPhase can also be controlled by tuning of varactor

loaded transmission lines [20] enabling a relatively

large bandwidth. To decrease circuit size, at low to

moderate frequencies, lumped-element-equivalent

structures as depicted in Figure 4 are favorable. Ac-

cording to [21], the design equations for the inductance

and the center capacitor of one P-low-pass segment

with equivalent characteristic length w between 0°

and 290° are given by

C5 2

tanaf2b

v # Z0

(3)

and

L5 2Z0sin 1f 2

v. (4)

By varying the varactor capacitance from the mini-

mum value Cvmin to the maximum value Cvmax, the

Figure 3. Active vector modulator operating at 5–5.5 GHz in 0.6 mm GaAs technology [11]. (a) Circuit schematic and (b) chip photo, chip size: 0.9 3 1.4 mm2.

5 pF

E50

50 fF

4 nH8 nH

10 pF

5 pF

Out

Vdc

5 nH5 nH

100 fF100 fF

200 fF

500 Ω

E50

E50

E50

7 pF

7 pF

7 pF

5 pF

5 pF

5 pF

In 1 k5

3 nH

Vgs, nominal =

0.25 V

850 Ω

360 fF 360 fF180 fF

7 nH7 nH

VHP

VLP

V0

500 Ω

500 Ω

500 Ω

500 Ω

500 Ω

Lowpass: –120°

Highpass: +120°

Direct: 0°

In

Out

VHP Vdc

(a)

(b)

E50

E50

850 Ω

850 Ω

Figure 4. Architecture of a distributed phase shifter based on low-pass filter sections [2].

C

Vc

2C

L

2C

L

2C

L

2C

L

2C

L

C

In Out

Rb

Cb

L

Figure 5. Phase control range versus equivalent characteristic line length and capacitance control range tv at a center operating frequency of one low-pass section.

180

160

140

120

100

80

60

40

20

00 –22.5 –45 –67.5 –90

Phase C

ontr

ol R

ange (

°)

Characteristic Line Length (°)

tv

1246810

102 October 2010

phase can be adjusted. As illustrated in Figure 5, the

higher the capacitance control ratio

tv5Cvmax

Cvmin

, (5)

and the higher the characteristic length of the equiva-

lent line, the higher the tuning range. The character-

istic length determines the difference in frequency

between the operating frequency and the resonance

frequency of the LC filters where maximum phase

variations are obtained. Unfortunately, higher induc-

tance values are required for increased characteristic

line lengths typically leading to raised loss. Hence, a

tradeoff between maximum phase control range and

minimum insertion loss has to be chosen. As a reason-

able rule of thumb, an equivalent characteristic length

of 55° per section is suggested. By cascading low-pass

sections, the phase control range can be multiplied.

Of course, at the same time the losses are multiplied

as well. At a typical tv of four, in ideal case, at least

nine sections are required to achieve a control range

of 360°. Considering the impairments in silicon, such

as fixed parasitic capacitances, resistive losses, process

variations and enabling a bandwidth of at least 20%, in

practice, a number of around 14 sections is required.

Such a circuit yielding 360° phase control plus

margin and a bandwidth of at least 4.2–6.2 GHz was

implemented in 0.18 mm CMOS technology [22]. The

control performance and the photo with chip area of

1 mm2 are depicted in Figure 6. At 5.2 GHz, the inser-

tion loss amounts to2 7.3 6 2 dB. A modified version,

optimized for low loss with six low-pass sections,

provides a phase control range of 195°, an insertion

loss of 3.6 6 1.5 dB, a P1dB of 12 dBm and a chip area

of 0.6 mm2 [23]. The tuning range per section can be

increased by tuning the inductances together with the

varactors. Active inductors as reported in [24] can be

used for this task.

Referring to [25], digital control of the capacitive

loading of the low-pass section can also be performed

using active distributed switches. The circuit has 4-bit

resolution at 11.6–12.6 GHz, 23.5 6 0.5 dB insertion

loss, 26.6 mW dc power and 1.7 mm2 chip area. A dif-

ferential version of a distributed phase shifter has

been presented in [26]. To compensate for losses,

distributed phase shifters can be combined with dis-

tributed amplifiers [27].

Coupler-Based Reflective-Type Phase ShiftersBy varying the element values of loads, the insertion

phase can be varied. However, typically such varia-

tions have a nondesired impact on the input and out-

put impedance matching of the circuit. A coupler can

be used to solve this problem. The signals are reflected

at the load terminals, which are isolated, by means of

the coupler, from the input and output nodes of the

phase shifter. The insertion phase variation of the cir-

cuit is proportional to the reflection coefficient at the

reflective loads given by

G5Zr2Z0

Zr1Z0, (6)

where Zr is the impedance of the reflective loads,

which in the ideal case is reactive, as discussed later.

Figure 6. Varactor-tuned equivalent-transmission-line phase shifter operating at 4–6 GHz with 360° phase control in 0.18 mm CMOS technology [22], Vc denotes the control voltage, w215/ 1S21 2 . (a) Gain and phase versus control voltage at 5.2 GHz and (b) chip photo, chip size 1.3 3 0.8 mm2.

–4

–5

–6–7

–8

–9

–10

–11–12

–13

–14–7 –6 –5 –4 –3 –2 –1 0

375

300

225150

75

0

–75–150

–225

–300

450

S21 (dB

)

ϕ 21

(°)

Vc /V

360°360

(a)

Out

Vc

In

(b)

S21 Sim

S21 Meas

ϕ21 Meas

ϕ21 Sim

Figure 7. Circuit topology of the reflective-type phase shifter using a branch-line coupler [2]. Phase control is obtained by varying the phase of the reflection factor G.

90°Coupler

In

Out Zr

Zr

Γ

Phase can also be controlled by tuning of varactor loaded transmission lines enabling a relatively large bandwidth.

October 2010 103

The reference impedance Z0 is assumed to be resistive

e.g., 50 V. These types of circuits are called reflective

type phase shifters. The phase control can be calcu-

lated from

Dw5 2 carctana 0 Im 1Zrmax 2 0Z0

b 2 arctana 0 Im 1Zrmin 2 0Z0

b d , (7)

where Zrmax and Zrmin denote the maximum and mini-

mum reflection impedances, respectively. Let us first

discuss the principle of the reflective-type phase

shifter based on a 90° branch-line coupler and two

equal reflective loads. The input signal is divided into

two parts. Each part is reflected at one of the reflective

loads. The loss of the circuit is given by

0 S21 0 5 20log 1 0G 0 2 . (8)

Referring to (8), resistive losses in the reflective

load have to be avoided to minimize the insertion

loss, which equals 0 dB if all elements are lossless.

Thus, reactive elements are favorable for the reflec-

tive loads.

If a varactor is used as reflective load, a phase con-

trol of up to 180° can be reached if the tuning range of

the varactor goes towards infinity. By resonating the

varactor with an inductor Lr around the center fre-

quency of the phase shifter, the phase control range

can be increased up to 360°. However, since the tun-

ing range of varactors is limited, the full 360° can-

not be achieved in practice. A further increase of the

phase control range is feasible by employing parallel

loads with different resonance frequencies associated

with different values of the varactor capacitances. In

Figure 8, simulated reflection coefficients are plotted

for four reflective load options. In these simulations,

typical values and losses for integrated implementa-

tions are considered. In this specific example, the Q

factors of the inductors and varactors at 5 GHz are

around 15 and 30, respectively. The closer the lines

are to the reference impedance (in our case 50 V),

the higher the insertion losses due to the fact that

a smaller part of the signal is reflected and a higher

part is absorbed in the reflective load. Both the inser-

tion losses and the associated ripple can be decreased

by adding a transformation network, which increases

the magnitude of the minimal reflection coefficient

and decreases the variations of the magnitude of the

reflection coefficient.

Figure 8. Simulated reflection coefficient of reflective loads including parasitics at 5.2 GHz. The varactor capacitance is varied from 0.18 pF to 0.7 pF [28].

Cv

Cv

Cv

Cv

CvCv

Cv

Cv

Lr = 2.3 nH

Lr 1 = 1.4 nHLr 1 = 1.4 nH

Cv Lr 2 = 4 nHLr 2 = 4 nHΔϕ ≈ 190°

Δϕ ≈ 410°

Δϕ ≈ 60°

Δϕ ≈ 410°

0.41 pF

1 nH

0.5 0.5

0.2 0.2

–0.2 –0.2

–0.5 –0.5

–1.0 –1.0

–2.0 –2.0

–5.0 –5.0

5.0 5.0

2.0 2.0

1.0 1.0

0.0 0.2 0.5 1.0 2.0 5.0 0.0 0.2 0.5 1.0 2.0 5.0

Compared to passive couplers, a much higher bandwidth can be achieved by the active circulators if broadband amplifier design techniques are used.

104 October 2010

A circuit operating between 5.1 and 5.7 GHz apply-

ing such reflective loads with two parallel resonators

and transformation network was reported in [28]. To

minimize the circuit dimensions, the coupler was

realized by lumped-elements. Information regarding

the design of such a lumped element coupler can be

found in [29]. Figure 9 depicts the measured perfor-

mance and the photo of the chip with size of 1 mm2

realized in 0.6 mm GaAs MESFET technology. At

5.25 GHz, the circuit has a 360° phase control range,

an insertion loss of 6. 4 6 3 dB loss, and a P1dB rang-

ing from 2 to 15 dBm.

A simple implementation with reflective loads con-

sisting of just one resonated path was demonstrated in

0.25 mm BiCMOS technology [30]. Between 14.9 and 15.7

GHz, the circuit has a phase control range of 120°, an

insertion loss of 5 6 1 dB and a chip area of 0.35 mm2.

The phase control range can be increased by vary-

ing the varactor and the inductor of the reflective loads

at the same time. This is possible by using active con-

trollable inductors. In [31], a corresponding circuit was

implemented in InP HEMT technology. Since external

hyperabrupt GaAs varactors are used, the circuit is not

fully integrated. The circuit yields a very low insertion

loss of 0.8 dB and a phase control of more than 225°

within a frequency range of 4.7–6.7 GHz, and has a

power consumption of 54 mW.

Circulator-Based Reflective-Type Phase ShiftersOptionally, as shown in Figure 10, a three-port circu-

lator is well suited to isolate the impedance changes

of the reflective load on the input impedance and

output impedance of the phase shifter. Based on

three amplifiers, active circulators can be designed

[32], [33]. Such active circulator-based reflective-type

phase shifters can be optimized to have a transmis-

sion of [34]

S21512 gm2

# Zr

11 gm2# Zr

, (9)

where gm2 represents the transconductance of the

amplifier driving the reflective load. Consequently,

the phase can be adjusted by tuning Zr, and in addi-

tion by tuning gm2. Since gm2 can be controlled within

a wide range, relative large phase control ranges can

be achieved. Moreover, high quality factor (Q) metal

insulator metal capacitors with fixed values can be

used for Zr, resulting in lower losses with respect to

varactors exhibiting a lower Q. Compared to passive

couplers, a much higher bandwidth can be achieved

by the active circulators if broadband amplifier design

techniques are used.

Such a circuit was reported in [34]. Implemented

in 0.18 mm BiCMOS and operating from 2.5 GHz to

10 GHz, the circuit provides a phase control range of

at least 90° and a gain of 0 6 0.5 dB at 1.8 V3 4.6 mA

Figure 9. Reflective-type phase shifter at 5.1–5.7 GHz with 360° phase control in 0.6 mm GaAs MESFET technology featuring a lumped-element coupler and reflective loads with two parallel resonators and transformation networks [28]. (a) Measured performance for four different samples and (b) chip photo, size: 0.85 3 1.1 mm2.

–25

–20

–15

–10

–5

0

5

10

15

20

25

–1 0 1 2 3 4 5 6–500

–400

–300

–200

–100

0

100

200

300

400

500

VC (V )

S21, S

11, S

22 (dB

)

ϕ 21 (

°)

Mode 1: Δϕ = 360°S21= –6.4 dB ± 3 dB

Mode 2: Δϕ = 90°S21= –2.8 dB ± 0.4 dB

Vc

In

Out

S21

ϕ21

(a)

(b)

S11

S22

Figure 10. Block diagram of a circulator-based phase shifter.

In Out

Zr

Γr

The need for DACs can be circumvented by using circuits that require “digital” control voltages with only two states: high or low.

October 2010 105

dc power, while consuming a chip area of 0.475 mm2

and a core circuit area as small as 0.023 mm2. The mea-

sured phase versus control voltage and the chip photo

are plotted in Figure 11.

Switched-Type Phase Shifters Until now, the discussions have been focused on phase

shifters with continuously adjustable phase by means

of analog control voltages. However, typically, in

practice, power consuming, costly DACs are required

to generate these analog control voltages. The need

for DACs can be circumvented by using circuits that

require “digital” control voltages with only two states:

high or low. The idea is to switch between different

phase paths. The individual phases can be realized by

means of delay lines, or better, because more compact,

by lumped high-pass/low-pass structures. An example

of a 4-bit phase shifter is illustrated in Figure 12. For

design guidelines of such high-pass/low-pass sec-

tions, the reader is referred to [2]. The phase resolution

of the circuit is determined by the least significant bit,

which in the example shown is 22.5° corresponding to

360°/bit-amount2. Eight (two times bit amount) digi-

tal control voltages and maybe also contact pads are

required for this task. The number of control lines and

contact pads can be reduced by applying a serializer

circuit. Generally, the required lines or inductors are

large. Thus, especially at low-to-moderate frequencies,

these kinds of phase shifters require a large circuit size,

thereby increasing the costs as well. The overall inser-

tion loss and chip size increase with bit resolution. As

a first-order estimation, we may estimate an insertion

loss per bit between 0.5 and 2 dB.

In [35], a 5-bit switched phase shifter covering a fre-

quency range of 17–21 GHz with 5 6 0.6 dB insertion

loss fabricated in 0.25-mm PHEMT technology was

reported. An insertion loss of 14.5 6 0.5 dB was mea-

sured for a 5-bit circuit in 0.18 mm CMOS at 12 GHz

[36]. A four-element phased array system in SiGe BiC-

MOS at 34–39 GHz using switched-type phase shifters

with 5-bit control was reported in [37].

Figure 11. Circulator-based phase shifter in 0.18 mm

BiCMOS [34]; (a) measured phase versus frequency and control voltage and (b) chip photo with 0.73 3 0.65 mm2 chip area and 0.023 mm2 core circuit area.

(b)

GND GND GND

GNDPTAT

GNDGNDGND

Active

Core

ONOP

VDD

VC

INP INN

–75

–50

–25

0

25

50

75

100

125

150

2.5 3

3.5 4

4.5 5

5.5 6

6.5 7

7.5 8

8.5 9

9.5 10

f (GHz)

(a)

ϕ 21 (

°)

VC

180°

–180°

90°

–90°

45°

–45°

22.5°

–22.5°

In

V180° V90° V45° V25°

V–180° V–90° V–45° V–25°

Out

Figure 12. Digital adjustable phase shifter with 4-bit resolution [2].

One important design parameter for phase shifters is the phase control range.

106 October 2010

TABLE 2. Comparison of performance of fully integrated phase shifters, TLPS: tuned equivalent transmission line phase shifter based on distributed low-pass sections, RTPS: reflective-type phase shifter VM: vector modulator, *at center/target frequency.

Ref.Phase Control [°]

Bandwidth [GHz]

Gain ± Ripple* [dB]

Power Supply [mW]

Control Voltages

Chip Area [mm2] Technology Comment

Vector modulators

[8] 360 5–6 1.5 n.a. 18 n.a. 1 0.18 mm CMOS Active

[9] 360 15–20 26 ± 4 <112 2 analog 0.72 0.18 mm CMOS Active

[10] 360 50–56 24.9 ± 0.9 23 n.a.analog

0.4 90 nm CMOS

Active

[11] 360 5.1–5.3 0.6 n.a. 10 3 analog 1.3 0.6 mm GaAs MESFET

Active

[12] 360 4.7–5.7 29 n.a. < 0 3 analog 1 0.6 mm GaAs MESFET

Passive

[19] 360 4.8–5.8 0 n.a. 9 3 analog 2.4 0.6 mm GaAs MESFET

Active

Distributed phase shifters

[21] 360 5–6 24 ± 1.7 < 0 1 analog 0.8 0.6 mm GaAs MESFET

LP sections, passive

[22] >360 4.2–6.2 27.3 ± 2 < 0 1 analog 1 0.18 mm BiCMOS

LP sections, passive

[23] >195 5.2–5.8 23.6 ± 1.5 < 0 1 analog 0.6 0.25 mm BiCMOS

LP sections, passive

[24] 96 2.6 ± n.a. 23.3 ± 0.5 31.5 2 analog 1 0.13 mm CMOS LP sections with active inductors

[26] 360 8 ± n.a. n.a. 170 1 analog 0.25 0.18 mm CMOS LP sections, active, differential

[25] 360 4 bit 11.6–12.6 3.5 ± 0.5 26.6 8 digital 1.8 0.18 mm CMOS Distributed active switches

[39] 360 3.5–4.5 20.3 ± 0.8 16–25 n.a. 0.24 0.18 mm CMOS HP with active inductors

[40] 96 3.875–4.125

22.8 ± 1.5 >0 n.a. 4 analog 0.3 0.18 mm CMOS HP with active inductor

Coupler-based reflective-type phase shifters

[29] 210 6.1–6.3 24.9 ± 0.9 < 0 1 analog 0.9 0.6 mm GaAs MESFET

Passive

[28] 360 5.15–5.7 26.4 ± 3 < 0 1 analog 0.9 0.6 mm GaAs MESFET

Passive

[30] >120 14.9–15.7 25 ± 1 < 0 1 analog 0.36 0.25 mm BiCMOS

Passive

Circulator-based reflective-type phase shifters

[34] >90 2.5–10 0 ± 0.5 5.6–8.3 1 analog 0.48 0.18 mm BiCMOS

Active

Switched-type phase shifters

[35] 3605 bit

17–21 –5 ± 0.6 < 0 10 digital

1.3 0.25 mm PHEMT

Passive

[36] 3605 bit

9–15 –14.5 ± 0.5 < 0 10 digital

4.3 0.18 mm CMOS Passive

October 2010 107

Phase Control Range/Frequency Multiplication One important design parameter for phase shifters is

the phase control range. For many applications a phase

control range of 360° is required. However, the phase

control range has to be traded off with other param-

eters, which makes the design challenging. Is it pos-

sible to multiply the phase control range in a relatively

simple way?

Consider that frequency multiplication divides the

signal period duration. Consequently, the phase con-

trol range is multiplied by the factor of the frequency

multiplication. Typically, the method of frequency/

phase multiplication relaxes the contradicting con-

straints regarding the phase control range and ampli-

tude variations. The drawback is the additional power

consumed by the frequency multipliers [38], which

should have good linearity and low noise to avoid sig-

nal deteriorations.

Hardware ComparisonThe measured results of state-of-the-art phase

shifter implementations are summarized in Table 2.

In this article, we have focused on implementa-

tions using standard integrated circuit technolo-

gies. Interesting properties can be achieved using

microelectromechanical system (MEMS) technol-

ogy featuring high performance passives. Regard-

ing MEMS technologies and corresponding low

loss phase shifter implementations, the reader is

referred to [41]–[43].

ConclusionsIn this article, the development constraints, architec-

tures, design guidelines, and hardware results of fully

integrated phase shifters have been presented. The

pros and cons of different approaches have been dis-

cussed. To make a long story short:

Active vector modulators have the capability to •

provide gain and even gain control, and do not

require any varactors. Hence, vector modulators

are well suited for a large range of applications

and chip technologies.

Tuned equivalent transmission line phase shift- •

ers are typically passive and do not yield gain

but have the advantage of a very large band-

width of several octaves due to the distributed

nature.

Coupler-based reflective type phase shifters do •

have a simple architecture but tend to have a rela-

tively small bandwidth if the coupler is realized

by lumped elements.

Active circulator-based reflective type phase •

shifters can be controlled by the transconduc-

tance of the amplifiers driving the reflective load

thereby enabling a wide bandwidth and rela-

tively low loss.

At the expense of additional dc power and noise, •

active inductors can be used to increase the phase

control range for the tuned equivalent transmis-

sion line and reflective type phase shifters.

Switched-type phase shifters tend to have a rela- •

tively large size, especially at low to moderate

frequencies but do not require a DAC for phase

control from the system point of view.

For some specific systems including wireless •

communication front ends, phase/frequency

multiplication techniques may be interesting to

boost the phase control range while yielding a

low-gain ripple.

Finally, the optimum choice of the phase shifter

architecture depends on the required specifications

and the available technologies.

AcknowledgmentThis work was in part funded by the European Com-

munity’s Seventh Framework Programme under Grant

Agreement No. 213952 (MIMAX).

References[1] S. K. Koul and B. Bhat, Microwave and Millimeter Wave Phase Shifters,

vol. II. Norwood, MA: Artech House, 1991.

[2] F. Ellinger, Radio Frequency Integrated Circuits and Technologies, 2nd

ed. Heidelberg, Germany: Springer-Verlag, 2008.

[3] F. Ellinger, U. Jörges, U. Mayer, and R. Eickhoff, “Analysis and

compensation of phase variations versus gain in amplifiers veri-

fied by SiGe HBT cascode RFIC,” IEEE Trans. Microwave Theory Tech., vol. 57, no. 8, pp. 1885–1894, Aug. 2009.

[4] F. Ellinger, C. Carta, L. Rodoni, G. von Büren, D. Barras, M.

Schmatz, and H. Jäckel, “BiCMOS variable gain LNA at C-band

with ultra low power consumption for WLAN,” in Proc. IEEE Int. Telecommunication Conf., Aug. 2004, CD-ROM.

[5] F. Ellinger and H. Jäckel, “Low cost BiCMOS variable Gain LNA at

Ku-band with ultra low power consumption for adaptive antenna

combining,” IEEE Trans. Microwave Theory Tech., vol. 52, no. 2, pp.

702–708, Feb. 2004.

[6] U. Mayer, F. Ellinger, and R. Eickhoff, “Analysis and Reduction of

Phase Variations of Variable Gain Amplifiers verified by CMOS Im-

plementation at C-Band,” IET J. Circuits, Devices Syst., to be published.

[7] K.-J. Koh and G. M. Rebeiz, “0.13 µm CMOS phase shifters for X-,

Ku-, and K-band phased arrays,” IEEE J. Solid-State Circuits, vol. 42,

pp. 2535–2545, Nov. 2007.

[8] N. Joram, U. Mayer, R. Eickhoff, and F. Ellinger, “Fully integrated

active CMOS vector modulator for 802.11a compliant diversity

transceivers,” in Proc. IEEE Int. Conf. Microwaves, Communications, Antennas and Electronic Systems, Nov. 2009, pp. 1–4.

[9] P.-S. Wu, H.-Y. Chang, M.-D. Tsai, T.-W. Huang, and H. Wang,

“New miniature 15–20-GHz continuous-phase/amplitude control

The optimum choice of the phase shifter architecture depends on the required specifications and the available technologies.

108 October 2010

MMICs using 0.18-µm CMOS technology,” IEEE Trans. Microwave Theory Tech., vol. 54, no. 1, pp. 10–19, Jan. 2006.

[10] P. B. Vadivelu, P. Sen, S. Sarkar, D. Dawn, S. Pinel, and J. Laskar,

“Integrated CMOS mm-wave phase shifters for single chip por-

table radar,” in IEEE MTT-S Int. Microwave Symp. Dig., June 2009,

pp. 565–568.

[11] F. Ellinger, U. Lott, and W. Bächtold, “An antenna diversity MMIC

vector modulator for HIPERLAN with low power consumption

and calibration capability,” IEEE Trans. Microwave Theory Tech., vol.

49, no. 5, pp. 964–969, May 2001.

[12] F. Ellinger, R. Vogt, and W. Bächtold, “A high yield, ultra small,

passive, vector modulator based phase shifter for smart antenna

combining at C-band,” in Proc. IEEE/CSIRO Asia Pacific Microwave Conf., Dec. 2000, pp. 794–797.

[13] F. Ellinger and W. Bächtold, “Novel principle for vector modula-

tor based phase shifter operating with one control voltage,” IEEE J. Solid-State Circuits, vol. 37, no. 10, pp. 1256–1259, Oct. 2002.

[14] G. Rebeiz and K. J. Koh, “Silicon RFICs for phased arrays,” IEEE Microwave Mag., vol. 10, no. 3, pp. 96–103, May 2009.

[15] T. Yu and G. M. Rebeiz, “A 22–24 GHz 4 element CMOS phase

array with on-chip coupling characterization,” IEEE J. Solid-State Circuits, vol. 43, no. 9, pp. 2134–2143, Sept. 2008.

[16] K. J. Koh and G. M. Rebeiz, “An X- and Ku-band 8-element

phased-array receiver in 0.18 µm SiGe BiCMOS technology,” IEEE J. Solid-State Circuits, vol. 43, no. 6, pp. 1360–1371, June 2008.

[17] K. J. Koh, J. W. May and G. M. Rebeiz, “A millimeter-wave (40–45

GHz) 16-element phased-array transmitter in 0.18 µm SiGe BiC-

MOS technology,” IEEE J. Solid-State Circuits, vol. 44, no. 5, pp.

1498–1509, May 2009.

[18] F. Ellinger and W. Bächtold, “Adaptive antenna receiver module

for WLAN at C-Band with low power consumption,” IEEE Micro-wave Wireless Compon. Lett., vol. 12, no. 9, pp. 348–350, Sept. 2002.

[19] F. Ellinger, R. Vogt, and W. Bächtold, “Calibratable adaptive an-

tenna combiner at 5.2 GHz with high yield for laptop interface

card,” IEEE Trans. Microwave Theory Tech. (Special Issue), vol. 48,

no. 12, pp. 2714–2720, Dec. 2000.

[20] A. S. Nagra and R. A. York, “Distributed analog phase shifters

with low insertion loss,” IEEE Trans. Microwave Theory Tech., vol.

47, no. 9, pp. 1705–1711, Sept. 1999.

[21] F. Ellinger, H. Jäckel, and W. Bächtold, “Varactor loaded transmis-

sion line phase shifter at C-band using lumped elements,” IEEE Trans. Microwave Theory Tech., vol. 51, no. 4, pp. 1135–1140, Apr. 2003.

[22] J. Wagner, U. Mayer, and F. Ellinger, “Passive transmission line

phase shifter at C-band in CMOS using lumped elements,” in Proc. 17th Int. Conf. IEEE Microwaves, Radar and Wireless Communications, May 2008, pp. 1–4.

[23] F. Ellinger, J. Wagner, U. Mayer, and R. Eickhoff, “Passive varactor

tuned equivalent transmission line phase shifter at C-band in 0.25

µm BiCMOS,” IET J. Circuits, Devices Syst., vol. 2, no. 4, pp. 355–360,

2008.

[24] M. A. Y. Abdalla, K. Phang, and G. V. Eleftheriades, “Printed and

integrated CMOS positive/negative refractive-index phase shift-

ers using tunable active inductors,” IEEE Trans. Microwave Theory Tech., vol. 55, no. 8, pp. 1611–1623, Aug. 2007.

[25] D.-W. Kang and S. Hong, “A 4-bit CMOS phase shifter using dis-

tributed active switches,” IEEE Trans. Microwave Theory Tech., vol.

55, no. 7, pp. 1476–1483, July 2007.

[26] S. Hamedi-Hagh and C. A. T. Salama, “A novel C-band CMOS

phase shifter for communication systems,” in Proc. Int. Symp. Cir-cuits and Systems, May 2003, vol. 2, pp. II-316–II-319.

[27] C. Lu and A.-V. H. Pham, “Development of multiband phase

shifters in 180 nm RF CMOS technology with active loss compen-

sation,” IEEE Trans. Microwave Theory Tech., vol. 54, no. 1, pp. 40–45,

Jan. 2006.

[28] F. Ellinger, R. Vogt, and W. Bächtold, “Ultra compact reflective

type phase shifter MMIC at C-band with 360° phase control range

for smart antenna combining,” IEEE J. Solid-State Circuits, vol. 37,

no. 4, pp. 481–486, Apr. 2002.

[29] F. Ellinger, R. Vogt, and W. Bächtold, “Compact reflective type phase

shifter MMIC for C-band using a lumped element coupler,” IEEE Trans. Microwave Theory Tech., vol. 49, no. 5, pp. 913–917, May 2001.

[30] F. Ellinger, “A 15 GHz reflective type phase shifter MMIC fab-

ricated on 0.25 µm SiGe BiCMOS technology,” in Proc. Int. Conf. Telecommunication, May 2005, CD-ROM.

[31] H. Hayashi, M. Muraguchi, Y. Umeda, and T. Enoki, “A high-Q

broad-band active inductor and its application to a low-loss analog

phase shifter,” IEEE Trans. Microwave Theory Tech., vol. 44, no. 12,

pp. 2369–2374, Dec. 1996.

[32] S. Tanaka, N. Shimomura, and K. Ohtake, “Active circulators-the

realization of circulators using transistors,” Proc. IEEE, vol. 53, no.

3, pp. 260–267, Mar. 1965.

[33] Y. Zeng and C. E. Saavedra, “An ultra-compact CMOS variable

phase shifter for 2-4-GHz ISM applications,” IEEE Trans. Microwave Theory Tech., vol. 56, no. 6, pp. 1349–1354, June 2008.

[34] U. Mayer, F. Ellinger, and R. Eickhoff, “Compact circulator based

phase shifter at C-band in BiCMOS,” in Proc. IEEE Radio Frequency Integrated Circuits Symp., June 2009, pp. 499–502.

[35] C. F. Campell and S. A. Brown, “A compact 5-bit phase-shifter

MMIC for K-band satellite communication systems,” IEEE Trans. Microwave Theory Tech., vol. 48, no. 12, pp. 2652–2656, Dec. 2000.

[36] D.-W. Kang, H. D. Lee, C.-H. Kim, and S. Hong, “Ku-band MMIC

phase shifter using a parallel resonator with 0.18-µm CMOS tech-

nology,” IEEE Trans. Microwave Theory Tech., vol. 54, no. 1, pp. 294–

301, Jan. 2006.

[37] D.-W. Kang, J.-G. Kim, B.-W. Min, and G. M. Rebeiz, “Single

and four-element Ka-band transmit/receive phased-array sili-

con RFICs with 5-bit amplitude and phase control,” IEEE Trans. Microwave Theory Tech., vol. 57, no. 12, pp. 3534–3543, Dec. 2009.

[38] F. Ellinger and W. Bächtold, “Compact and low power consuming

frequency/phase multiplier MMICs for wireless LAN at S-band

and C-band,” IEE Proc. Devices, Circuits Syst., vol. 150, no. 3, pp.

199–204, June 2003.

[39] L.-H. Lu and Y.-T. Liao, “A 4-GHz phase shifter MMIC in 0.18-µm

CMOS,” IEEE Microwave Wireless Compon. Lett., vol. 15, no. 10, pp.

694–696, Oct. 2005.

[40] M. Abdalla, K. Phang, and G.-V. Eleftheriades, “A bi-directional

electronically tunable CMOS phase shifter using the high-pass

topology,” in IEEE MTT-S Int. Microwave Symp. Dig., June 2007, pp.

2173–2176.

[41] G. Rebeiz, K. Entesari, I. Reines, S.-J. Park, M. El-tanani, A.

Grichener, and A. Brown, “Tuning in to RF MEMS,” IEEE Micro-wave Mag., vol. 10, no. 6, pp. 55–72, Oct. 2009.

[42] N. S. Barker and G. M. Rebeiz, “Optimization of distributed

MEMS transmission-line phase shifters-U-band and W-band de-

signs,” IEEE Trans. Microwave Theory Tech., vol. 48, pp. 1957–1966,

Nov. 2000.

[43] K. Topalli, O. A. Civi, S. Demir, S. Koc, and T. Akin, “A mono-

lithic phased array using 3-bit distributed RF MEMS phase shift-

ers,” IEEE Trans. Microwave Theory Tech., vol. 56, no. 2, pp. 270–277,

Feb. 2008.